Fine tuning the classifier for new schemes

[GautamR-Samagra]

SAFAL-SUGAM Schemes Mapping.xlsx Above is a dataset for schemes in Odisha.

Below are some Schemes for Rajasthan : 'Diggi' 'Jalnidhi' 'PM-Kisan' 'PM-Kissan' 'Jalhauz'

• Create a single list of distinct schemes containing name like ['Scheme A','Scheme B'] etc

For each of the schemes, create questions like :

• {Scheme A}
• Where can I apply for {Scheme A}
• Tell me about {Scheme A}
• Who is eligible for {Scheme A}
• How to apply for {Scheme A}

Do not use the word scheme in these questions (classifier works well if the word scheme is mentioned. retrain the bert-base-uncased and the distilbert classifier (in that order) for this additional dataset and log the eval scores.

---

some other additions :	#	Query	Ideal Classifier	Classified as
1	Tell me about PM kisan	Scheme	Weather
2	Dealers in undoo	POP- dealer	Non-Agri
3	Where can i get machinery in barmer	POP- dealer	Non-Agri
4	Tell me about Diggi scheme	Scheme	Pest
5	Tell me about solar pump	Agri	Seed
6	Tell me about solar pump set	Scheme	Seed
7	Tell me about PM Kusum	Scheme	Non-Agri
8	Tell me about shade net scheme	Scheme	Non-Agri

[GautamR-Samagra]

@Jiya126